Advances in physical reservoir computing, a technology that makes sense of brain signals, could allow robots to be taught to think like humans, researchers said Wednesday. The technology was used to train robots to navigate through a maze by electrically stimulating human brain nerve cells connected to the machine, the scientists, from the University of Tokyo, wrote in an article in Applied Physics Letters.
These nerve cells, or neurons, were grown from living cells and acted as the physical “reservoir” for the computer to generate coherent signals. The signals, which are homeostatic, or living, effectively guided the robot through the maze, according to the researchers. Whenever the robot veered in the wrong direction or faced the wrong way, the neurons in the cell culture were stimulated by an electric impulse. Throughout trials, the robot was continually fed signals interrupted by the electrical impulses, which acted as disturbance signals, until it had successfully navigated the maze.
These findings suggest goal-directed behaviour can be taught to robots without any additional learning, according to the researchers. The robot could not see the environment or obtain other sensory information, so it was entirely dependent on the electrical trial-and-error impulses, the researchers said. “I, myself, was inspired by our experiments to hypothesize that intelligence in a living system emerges from a mechanism extracting a coherent output from a disorganized state,” study co-author Hirokazu Takahashi said in a press release.
Using this principle, intelligent task-solving abilities can be learned using physical reservoir computers to extract chaotic neuronal signals and deliver homeostatic or disturbance signals, said Takahashi, an associate professor of mechano-informatics. In doing so, the computer creates a reservoir that understands how to solve the task. Using physical reservoir computing in this way will contribute to a better understanding of the brain’s mechanisms and may lead to the novel development of a neuromorphic computer, according to Takahashi and his colleagues.
“A brain of [an] elementary school kid is unable to solve mathematical problems in a college admission exam, possibly because the dynamics of the brain or their ‘physical reservoir computer’ is not rich enough,” Takahashi said.
“Task-solving ability is determined by how rich a repertoire of spatiotemporal patterns the network can generate,” he said.